Promote audited AXQuant v2 artifact at stable repository name

Preserve backward-compatible repository identity; v1 remains tagged legacy-pre-v2 and this exact revision is tagged v2 after verification.

README.md

@@ -15,6 +15,7 @@ tags:
 - mistral-devstral
 - 4bit
 - 4-bit
+- v2
 ---
 
 # AX-Devstral-Small-2505-MLX-AXQ-4bit
@@ -26,17 +27,21 @@ the BF16 source model. The language path is quantized under AXQuant protection f
 > artifact-integrity records, but it does not publish measured quality, long-context, kernel-speed,
 > or MTP-speed evidence. Do not interpret the AXQ product label as a benchmark claim.
 
+> **Stable-name v2.** `main` serves the audited v2 artifact for backward compatibility. The same revision is tagged `v2`; the replaced artifact remains recoverable at `legacy-pre-v2`.
+
+
 ## Model details
 
 | Property | Value |
 | --- | --- |
-| Base model | [mistralai/Devstral-Small-2505](https://huggingface.co/mistralai/Devstral-Small-2505) |
-| Source revision | `unrecorded` |
+| Base model | [mistralai/Devstral-Small-2505](https://huggingface.co/mistralai/Devstral-Small-2505/tree/c2a9d81a2989af566682b4cecc828c84556076c5) |
+| Source revision | `c2a9d81a2989af566682b4cecc828c84556076c5` |
 | Product family | `mistral-devstral` |
 | Source architecture | `MistralForCausalLM` (dense); text path optimized |
 | Main-model parameters | 23.57B logical parameters |
-| Quantizer | AXQuant `1.0.0` |
+| Quantizer | AXQuant `1.2.0` |
 | Hub budget class | `4bit` |
+| Artifact edition | `v2` |
 | AXQuant base precision class | `4bit` |
 | Planned storage-adjusted BPW | 4.9499 |
 | Measured main-model BPW | 4.9500 |
@@ -44,8 +49,8 @@ the BF16 source model. The language path is quantized under AXQuant protection f
 | Safetensors weight size | 14.59 GB |
 | Approximate complete download | 14.60 GB |
 | Configured maximum context | 131,072 tokens; practical limits depend on unified memory |
-| Primary runtime | AX Engine, compatibility level A |
-| Compatible runtime | MLX-LM standard text inference, compatibility level B |
+| MLX-LM compatibility | Standard text inference, compatibility level B |
+| AX Engine native execution | Not established; no validated native manifest is included |
 | MTP present | `False` |
 | Vision sidecar present | `False` |
 
@@ -63,7 +68,7 @@ protected models.
 | Sibling | Intended trade-off |
 | --- | --- |
 | [4bit sibling](https://huggingface.co/AutomatosX/AX-Devstral-Small-2505-MLX-AXQ-4bit) | Lower-storage AXQ budget; check its exact BPW |
-| [6bit sibling](https://huggingface.co/AutomatosX/AX-Devstral-Small-2505-MLX-AXQ-6bit) | Higher average precision near a 6-BPW budget |
+| [6bit sibling](https://huggingface.co/AutomatosX/AX-Devstral-Small-2505-MLX-AXQ-6bit) | Higher average precision near the 6-BPW budget |
 
 See the [AutomatosX MLX model catalog](https://huggingface.co/collections/AutomatosX/automatosx-mlx-model-catalog)
 for related MLX and OptiQ alternatives.
@@ -94,19 +99,13 @@ metadata and optional sidecars (`vision.safetensors`, `mtp.safetensors`); this c
 does not establish MTP acceleration or vision-language quality. The artifact records MLX
 `0.32.0` and MLX-LM `0.31.3` from conversion.
 
-## Serve with AX Engine
+## AX Engine status
 
-After installing [AX Engine](https://github.com/defai-digital/ax-engine), download the complete
-repository and serve the local directory:
-
-```bash
-ax-engine serve ./AX-Devstral-Small-2505-MLX-AXQ-4bit --port 31418
-```
-
-AX Engine is the authority for the AXQ runtime contract.
-This development package does not claim runtime speedups until identical-checkpoint benchmarks are
-published. The artifact records AX Engine version `not recorded`. Native
-`model-manifest.json` status: included as `model-manifest.json`.
+This package does **not** include a validated native `model-manifest.json`, so AX Engine execution
+is not established by this release. The AX Engine fields in `axquant_runtime.json` describe the
+intended compatibility contract, not observed runtime evidence. Use the MLX-LM path above for
+standard text/backbone inference. The artifact records AX Engine version
+`not recorded`, but version discovery alone is not a runtime check.
 
 ## Quantization layout
 
@@ -133,7 +132,7 @@ establish MTP acceleration or vision-language quality.
 | Planning evidence | `architecture_prior` |
 | Calibration | none; the allocation is based on architecture priors |
 | Quantizer execution | 281/281 recorded module conversions succeeded; 0 fallbacks |
-| AX Engine native manifest | included as `model-manifest.json` |
+| AX Engine native manifest | not included |
 | Quality versus BF16 or uniform baselines | Not published; no quality-retention claim |
 | MTP acceptance and speed | not measured; no MTP speedup claim |
 | AX Engine kernel evidence | `unmeasured` |
@@ -149,6 +148,8 @@ establish MTP acceleration or vision-language quality.
 - Architecture-prior allocation is not measured sensitivity. It must not be presented as measured
   model quality.
 - The configured context window can require substantially more memory as the KV cache grows.
+- AX Engine execution is not established because this package has no validated native manifest.
+
 - Upstream capabilities, limitations, biases, and responsible-use guidance still apply.
 
 ## Provenance and audit files
@@ -158,8 +159,7 @@ establish MTP acceleration or vision-language quality.
 - [`axquant_plan.json`](axquant_plan.json): per-tensor precision decisions and planning evidence.
 - [`axquant_quantizer_execution.json`](axquant_quantizer_execution.json): conversion coverage and
   fallback records.
-- [`axquant_runtime.json`](axquant_runtime.json): AX Engine and MLX-LM compatibility contract.
-- [`model-manifest.json`](model-manifest.json): AX Engine native tensor manifest.
+- [`axquant_runtime.json`](axquant_runtime.json): declared AX Engine and MLX-LM compatibility metadata; runtime checks remain separate evidence.
 
 All published provenance uses repository-relative paths. Local source paths are stripped before
 publication. The checkpoint was converted from BF16 rather than re-quantized from an OptiQ
@@ -169,5 +169,5 @@ identical BPW or quality.
 ## License
 
 The checkpoint follows the upstream model license where applicable (often Apache License 2.0). See
-the [mistralai/Devstral-Small-2505 model card](https://huggingface.co/mistralai/Devstral-Small-2505) for license terms, model
+the [mistralai/Devstral-Small-2505 model card](https://huggingface.co/mistralai/Devstral-Small-2505/tree/c2a9d81a2989af566682b4cecc828c84556076c5) for license terms, model
 limitations, and responsible-use guidance.

axquant_manifest.json

@@ -1,28 +1,28 @@
 {
-  "axquant_version": "1.0.0",
+  "axquant_version": "1.2.0",
   "calibration": null,
-  "created_at": "2026-08-02T18:31:02.216726Z",
+  "created_at": "2026-08-05T08:23:55.481291Z",
   "effective_bpw": 4.949927508451917,
   "files": [
     {
       "path": "README.md",
-      "sha256": "99727356000056a46b9405941aadc9ddc9bc15baca18059b1f07553ec35ec358",
-      "size_bytes": 7465
+      "sha256": "f944a3bed85ad156433c73ce4227cc8dd7e5de07a21b281f132ff8b5bf5c0dca",
+      "size_bytes": 7796
     },
     {
       "path": "axquant_plan.json",
-      "sha256": "919f9565070912d0f01efe80c4673550f41383fe5226561b211e4d1c1bafdd78",
-      "size_bytes": 337327
+      "sha256": "6f90c1d59273dd8c482a7b14bccc90b8f7947e244ec71a572fd30ee71fe015c1",
+      "size_bytes": 337379
     },
     {
       "path": "axquant_quantizer_execution.json",
-      "sha256": "1c8499ca87fb7ec2c371ac4876e86a7faa58c3038a70e7e40bba9a28e374a424",
+      "sha256": "226b42e3e7a08581a1c8128c52ed319d5263807fafe71b47e356f324933a8b61",
       "size_bytes": 68529
     },
     {
       "path": "axquant_runtime.json",
-      "sha256": "a0163c63a8fab3e93ef6c027e4a509fcd380267b4252b4137cb8213c9695917b",
-      "size_bytes": 1643
+      "sha256": "3c218a96a5a8d62d687f22d0a91985dc9164f5c8b15770573d0eaeb3e5811897",
+      "size_bytes": 1628
     },
     {
       "path": "chat_template.jinja",
@@ -41,7 +41,7 @@
     },
     {
       "path": "model-00001-of-00003.safetensors",
-      "sha256": "d0a2e91e9d8ca5322df7d3d184ada4f68f9c5a4772d551d1fc9d8e5a1135c81b",
+      "sha256": "b6580465a51a48c0d4541aaaf3017ace61f17e7bee36c49a5e3eb5e53ce0bf27",
       "size_bytes": 5308086419
     },
     {
@@ -54,11 +54,6 @@
       "sha256": "9060d2f447d0203a0459a36cba4ef510e3c4eebb4cae92c8228129bc123407e5",
       "size_bytes": 3962522975
     },
-    {
-      "path": "model-manifest.json",
-      "sha256": "0ebe4e82e362ae95227b6d92490009672f6d8d6ca1ade22c588cdd9e3bb95f8f",
-      "size_bytes": 150358
-    },
     {
       "path": "model.safetensors.index.json",
       "sha256": "49f2e5cc33c02b3040dbff542839e016af6fe9b0ba8aaeea67e31df919ac715a",
@@ -98,7 +93,7 @@
   },
   "mtp_present": false,
   "mtp_weight_file_size_bytes": 0,
-  "plan_sha256": "1064d8412710fbf2af1fd2d7d9ffb918415a396790d4ce5916facf40312a5dd7",
+  "plan_sha256": "7bd8665657133cce8bda462f969b954f466519dbff00600f596688ed5e84888f",
   "profile": "general",
   "protected_weight_file_size_bytes": 0,
   "quantizer": "axquant",
@@ -114,7 +109,7 @@
       {
         "compatibility_level": "B",
         "manifest": "config.json",
-        "mtp_support": "runtime-dependent",
+        "mtp_support": "none",
         "name": "mlx-lm",
         "notes": [
           "Standard backbone inference is the compatibility target.",
@@ -125,7 +120,7 @@
         "support_level": "standard-inference"
       }
     ],
-    "created_at": "2026-08-02T18:30:54.875389Z",
+    "created_at": "2026-08-05T08:23:50.819054Z",
     "kv_cache": null,
     "memory_policy": {
       "kv_cache_precision": "runtime-default",
@@ -149,7 +144,7 @@
     "primary_runtime": {
       "compatibility_level": "A",
       "manifest": "model-manifest.json",
-      "mtp_support": "native",
+      "mtp_support": "none",
       "name": "ax-engine",
       "notes": [
         "Runtime claims require a passing AX Engine doctor and benchmark report."
@@ -163,7 +158,7 @@
   "schema_version": "axquant.artifact.v2",
   "software_versions": {
     "ax_engine": null,
-    "axquant": "1.0.0",
+    "axquant": "1.2.0",
     "mlx": "0.32.0",
     "mlx_lm": "0.31.3",
     "pydantic": "2.13.4",
@@ -175,7 +170,7 @@
     "format": "mlx",
     "local_path": null,
     "model_id": "mistralai/Devstral-Small-2505",
-    "revision": null
+    "revision": "c2a9d81a2989af566682b4cecc828c84556076c5"
   },
   "target_class": "4bit",
   "weight_distribution": {

axquant_plan.json

@@ -1,5 +1,5 @@
 {
-  "analysis_sha256": "21c80c98be1576ed3f52cd6be881bb9cd7e67ef50a3e830e23d155ceaebfacee",
+  "analysis_sha256": "9c2c4dbc8887f3c2b65a95370cbef6855c3bef3054a0e0ea47e1380e279ad7bb",
   "architecture_profile": {
     "adapter_id": "mistral-devstral-dense-v1",
     "config_model_type": "mistral",
@@ -10564,7 +10564,7 @@
     "minimum_mtp_speedup": 1.2,
     "minimum_quality_retention": 0.98
   },
-  "created_at": "2026-08-02T18:30:28.432333Z",
+  "created_at": "2026-08-05T08:22:48.557763Z",
   "effective_bpw": 4.949927508451917,
   "evidence_kind": "architecture_prior",
   "global_validation_required": true,
@@ -10589,6 +10589,7 @@
       "affine",
       "awq",
       "dwq",
+      "gptq",
       "bf16"
     ]
   },
@@ -10626,7 +10627,7 @@
   "schema_version": "axquant.plan.v1",
   "software_versions": {
     "ax_engine": null,
-    "axquant": "1.0.0",
+    "axquant": "1.2.0",
     "mlx": "0.32.0",
     "mlx_lm": "0.31.3",
     "pydantic": "2.13.4",
@@ -10638,7 +10639,7 @@
     "format": "mlx",
     "local_path": null,
     "model_id": "mistralai/Devstral-Small-2505",
-    "revision": null
+    "revision": "c2a9d81a2989af566682b4cecc828c84556076c5"
   },
   "status": "planned",
   "target_bpw": 4.95,

axquant_quantizer_execution.json

@@ -1,6 +1,6 @@
 {
-  "created_at": "2026-08-02T18:30:54.870992Z",
-  "plan_sha256": "1064d8412710fbf2af1fd2d7d9ffb918415a396790d4ce5916facf40312a5dd7",
+  "created_at": "2026-08-05T08:23:50.814845Z",
+  "plan_sha256": "7bd8665657133cce8bda462f969b954f466519dbff00600f596688ed5e84888f",
   "records": [
     {
       "bits": 8,

axquant_runtime.json

@@ -10,7 +10,7 @@
     {
       "compatibility_level": "B",
       "manifest": "config.json",
-      "mtp_support": "runtime-dependent",
+      "mtp_support": "none",
       "name": "mlx-lm",
       "notes": [
         "Standard backbone inference is the compatibility target.",
@@ -21,7 +21,7 @@
       "support_level": "standard-inference"
     }
   ],
-  "created_at": "2026-08-02T18:30:54.875389Z",
+  "created_at": "2026-08-05T08:23:50.819054Z",
   "kv_cache": null,
   "memory_policy": {
     "kv_cache_precision": "runtime-default",
@@ -45,7 +45,7 @@
   "primary_runtime": {
     "compatibility_level": "A",
     "manifest": "model-manifest.json",
-    "mtp_support": "native",
+    "mtp_support": "none",
     "name": "ax-engine",
     "notes": [
       "Runtime claims require a passing AX Engine doctor and benchmark report."

model-00001-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d0a2e91e9d8ca5322df7d3d184ada4f68f9c5a4772d551d1fc9d8e5a1135c81b
+oid sha256:b6580465a51a48c0d4541aaaf3017ace61f17e7bee36c49a5e3eb5e53ce0bf27
 size 5308086419